Differences in the predictive capability for functional impairment, cognitive decline and mortality of different frailty tools: A longitudinal cohort study.

Background And Objectives: There are multiple frailty detection tools, but they have not been specifically developed for the institutionalised population. The aim of this study is to ascertain at 3-year follow-up which tool predicts functional impairment and mortality most precisely.

Methods: Longitudinal cohort study with 110 patients in Pamplona (Navarra)>65 years. Four frailty tools were applied (Fried Criteria, Rockwood Frailty Scale, FRAIL-NH and Imputed Fried Frailty Criteria). The power of the association between the scales and the results was assessed by linear regression and Cox's analyses.

Results: 46.5% of the sample died during time to follow-up, 68% of whom died in their nursing home, with 43-month mean survival. Of the studied population, 71.3% showed disability at 3 years, especially the frail subjects. The robust patients had longer hospitalizations (m=3.4 days) than the frail. Imputed Fried and FRAIL-NH found statistically significant differences between groups for the variables studied. Imputed Fried Frailty Criteria showed a significant HR of death for the frail subjects (HR=3.3).

Conclusions: The Imputed Fried and FRAIL-NH tools showed a higher predictive capability for functional and cognitive decline, but only the Imputed Fried Frailty Criteria found a significant relationship between frailty and mortality.